
import matplotlib.pyplot as plt
import numpy as np
import os
import PIL
import tensorflow as tf
import pandas as pd
from tensorflow import keras
from tensorflow.keras import layers
from tensorflow.keras.models import Sequential
import imageio
from PIL import Image
import ipyplot
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.naive_bayes import GaussianNB
from sklearn.ensemble import RandomForestClassifier
from xgboost import XGBClassifier
from sklearn.metrics import classification_report
from sklearn.metrics import precision_recall_fscore_support as score, precision_score, recall_score, f1_score
import pathlib
import io
import os
import requests
import pathlib
import gzip
import imageio
import pandas as pd


import pathlib
data_dir = "hpa-single-cell-image-classification/train"
data_dir = pathlib.Path(data_dir)


image_count = os.listdir(data_dir)
len(image_count)

87224


df_train = pd.read_csv("hpa-single-cell-image-classification/train.csv")
df_train.head()


df_samp_sub = pd.read_csv("hpa-single-cell-image-classification/sample_submission.csv")
df_samp_sub.head()


train_images = "hpa-single-cell-image-classification/train"
train_images = os.listdir(train_images)
#train_images


len(df_train), len(train_images), len(train_images)/len(df_train)

(21806, 87224, 4.0)


df_train['num_classes'] = df_train['Label'].apply(lambda r: len(r.split('|')))
df_train['num_classes'].value_counts().plot.bar(title='Examples with multiple labels', xlabel='number of labels per example', ylabel='# train examples')
plt.show()


path = "hpa-single-cell-image-classification"
mt = [path+'/train/'+(id+'_red.png') for id in df_train.ID.values]
er = [path+'/train/'+(id+'_yellow.png') for id in df_train.ID.values]
nu = [path+'/train/'+(id+'_blue.png') for id in df_train.ID.values]
pr = [path+'/train/'+(id+'_green.png') for id in df_train.ID.values]
images = [mt, er, nu, pr]
titles = ['microtubules', 'endoplasmic reticulum', 'nucleus', 'protein of interest']


fig, ax = plt.subplots(3,4, figsize=(20,12))
for j in range(3):
    for i in range(4):
        img = plt.imread(images[i][j])
        if j == 0: ax[j,i].set_title(titles[i])
        ax[j,i].imshow(img)
        ax[j,i].axis('off')
plt.show()


labels = [str(i) for i in range(19)]
class_images = []
for label in labels:
    r_img = df_train[df_train.Label == label].reset_index(drop=True).ID.loc[0] + '_red.png'
    y_img = df_train[df_train.Label == label].reset_index(drop=True).ID.loc[0] + '_yellow.png'
    b_img = df_train[df_train.Label == label].reset_index(drop=True).ID.loc[0] + '_blue.png'
    r = imageio.imread(path+'/train/'+r_img)
    y = imageio.imread(path+'/train/'+y_img)
    b = imageio.imread(path+'/train/'+b_img)
    rgb = np.dstack((r,y,b))
    #print(r.shape)
    class_images.append(PIL.Image.frombytes('RGB',r.shape,rgb))

codes = [
'0. Nucleoplasm',
'1. Nuclear membrane',
'2. Nucleoli',
'3. Nucleoli fibrillar center',
'4. Nuclear speckles',
'5. Nuclear bodies',
'6. Endoplasmic reticulum',
'7. Golgi apparatus',
'8. Intermediate filaments',
'9. Actin filaments',
'10. Microtubules',
'11. Mitotic spindle',
'12. Centrosome',
'13. Plasma membrane',
'14. Mitochondria',
'15. Aggresome',
'16. Cytosol',
'17. Vesicles and punctate cytosolic patterns',
'18. Negative'
]

#ipyplot.plot_images(images=class_images, labels=codes, max_images=19, img_width=300)
ipyplot.plot_images(images=class_images, labels=codes, max_images=19)


# # per label 
# labels = [str(i) for i in range(19)]
# code = [
# '00_Nucleoplasm',
# '01_Nuclear_membrane',
# '02_Nucleoli',
# '03_Nucleoli_fibrillar_center',
# '04_Nuclear_speckles',
# '05_Nuclear_bodies',
# '06_Endoplasmic_reticulum',
# '07_Golgi_apparatus',
# '08_Intermediate_filaments',
# '09_Actin_filaments',
# '10_Microtubules',
# '11_Mitotic_spindle',
# '12_Centrosome',
# '13_Plasma_membrane',
# '14_Mitochondria',
# '15_Aggresome',
# '16_Cytosol',
# '17_Vesicles_and_punctate_cytosolic_patterns',
# '18_Negative'
# ]

# for label in labels:
#     img_counter = len(df_train[df_train.Label == label])
#     print("Generating Images for "+code[int(label)] )
#     for i in range(img_counter):
#         try:
#             r_img = df_train[df_train.Label == label].reset_index(drop=True).ID.loc[i] + '_red.png'
#             y_img = df_train[df_train.Label == label].reset_index(drop=True).ID.loc[i] + '_yellow.png'
#             b_img = df_train[df_train.Label == label].reset_index(drop=True).ID.loc[i] + '_blue.png'
#             name = df_train[df_train.Label == label].reset_index(drop=True).ID.loc[i]

#             r = imageio.imread(path+'/train/'+r_img)
#             y = imageio.imread(path+'/train/'+y_img)
#             b = imageio.imread(path+'/train/'+b_img)
#             rgb = np.dstack((r,y,b))
            
#             image_name = PIL.Image.frombytes('RGB',r.shape,rgb)
#             resized_image = image_name.resize((180,180))

#             if not os.path.exists("data/"+code[int(label)]):
#                 os.makedirs("data/"+code[int(label)])

#             resized_image.save("data/"+code[int(label)]+"/"+ name + ".png")
#         except:
#             print("Unable to save image with this label: "+ code[int(label)])
    
#     print("Completed Generating Images for "+code[int(label)])


batch_size = 32
img_height = 180
img_width = 180


train_ds = tf.keras.preprocessing.image_dataset_from_directory(
 "./data",
  validation_split=0.2,
  subset="training",
  seed=123,
  image_size=(img_height, img_width),
  batch_size=batch_size)

Found 10412 files belonging to 19 classes.
Using 8330 files for training.


val_ds = tf.keras.preprocessing.image_dataset_from_directory(
  "./data",
  validation_split=0.2,
  subset="validation",
  seed=123,
  image_size=(img_height, img_width),
  batch_size=batch_size)

Found 10412 files belonging to 19 classes.
Using 2082 files for validation.


x_train = list()
y_train = list()
for raw_data,labels in train_ds:
    y_train.extend(np.array(labels))
   # print(labels)
    for image_data_per in raw_data:
        x_train.append(np.array(image_data_per))
x_train = np.array(x_train).reshape(-1,(180*180*3))
print(x_train.shape)
y_train = np.array(y_train)
print(y_train.shape)

(8330, 97200)
(8330,)


x_test = list()
y_test = list()
for raw_data,labels in val_ds:
    y_test.extend(np.array(labels))
    #print(labels)
    for image_data_per in raw_data:
        x_test.append(np.array(image_data_per))
x_test = np.array(x_test).reshape(-1,(180*180*3))
print(x_test.shape)
y_test = np.array(y_test)
print(y_test.shape)


x_train = x_train[0:1000]
y_train = y_train[0:1000]
x_test = x_test[0:100]
y_test = y_test[0:100]


import sklearn.metrics as metrics
import seaborn as sns
"""
def plot_multiclass_roc(clf, X_test, y_test, n_classes, figsize=(17, 6)):
    y_score = clf.decision_function(X_test)

    # structures
    fpr = dict()
    tpr = dict()
    roc_auc = dict()
    label_ = dict(code)
    
    # calculate dummies once
    y_test_dummies = pd.get_dummies(y_test, drop_first=False).values
    for i in range(n_classes):
        fpr[i], tpr[i], _ = metrics.roc_curve(y_test_dummies[:, i], y_score[:, i])
        roc_auc[i] = metrics.auc(fpr[i], tpr[i])

    # roc for each class
    fig, ax = plt.subplots(figsize=figsize)
    ax.plot([0, 1], [0, 1], 'k--')
    ax.set_xlim([0.0, 1.0])
    ax.set_ylim([0.0, 1.05])
    ax.set_xlabel('False Positive Rate')
    ax.set_ylabel('True Positive Rate')
    ax.set_title('Receiver operating characteristic example')
    for i in range(n_classes):
        ax.plot(fpr[i], tpr[i], label='ROC curve (AUC = %0.2f) for label %i' % (roc_auc[i], i))
    ax.legend(loc="best")
    ax.grid(alpha=.4)
    sns.despine()
    plt.show()
"""

'\ndef plot_multiclass_roc(clf, X_test, y_test, n_classes, figsize=(17, 6)):\n    y_score = clf.decision_function(X_test)\n\n    # structures\n    fpr = dict()\n    tpr = dict()\n    roc_auc = dict()\n    label_ = dict(code)\n    \n    # calculate dummies once\n    y_test_dummies = pd.get_dummies(y_test, drop_first=False).values\n    for i in range(n_classes):\n        fpr[i], tpr[i], _ = metrics.roc_curve(y_test_dummies[:, i], y_score[:, i])\n        roc_auc[i] = metrics.auc(fpr[i], tpr[i])\n\n    # roc for each class\n    fig, ax = plt.subplots(figsize=figsize)\n    ax.plot([0, 1], [0, 1], \'k--\')\n    ax.set_xlim([0.0, 1.0])\n    ax.set_ylim([0.0, 1.05])\n    ax.set_xlabel(\'False Positive Rate\')\n    ax.set_ylabel(\'True Positive Rate\')\n    ax.set_title(\'Receiver operating characteristic example\')\n    for i in range(n_classes):\n        ax.plot(fpr[i], tpr[i], label=\'ROC curve (AUC = %0.2f) for label %i\' % (roc_auc[i], i))\n    ax.legend(loc="best")\n    ax.grid(alpha=.4)\n    sns.despine()\n    plt.show()\n'


def plot_multiclass_roc_advanced(clf, X_test, y_test, n_classes, figsize=(17, 6)):
    y_score = clf.predict_proba(X_test)

    # structures
    fpr = dict()
    tpr = dict()
    roc_auc = dict()
    # calculate dummies once
    y_test_dummies = pd.get_dummies(y_test, drop_first=False).values
    for i in range(n_classes):
        fpr[i], tpr[i], _ = metrics.roc_curve(y_test_dummies[:, i], y_score[:, i])
        roc_auc[i] = metrics.auc(fpr[i], tpr[i])

    # roc for each class
    fig, ax = plt.subplots(figsize=figsize)
    ax.plot([0, 1], [0, 1], 'k--')
    ax.set_xlim([0.0, 1.0])
    ax.set_ylim([0.0, 1.05])
    ax.set_xlabel('False Positive Rate')
    ax.set_ylabel('True Positive Rate')
    ax.set_title('Receiver operating characteristic example')
    for i in range(n_classes):
        ax.plot(fpr[i], tpr[i], label='ROC curve (AUC = %0.2f) for label %i' % (roc_auc[i], i))
    ax.legend(loc="best")
    ax.grid(alpha=.4)
    sns.despine()
    plt.show()


pipe = make_pipeline(StandardScaler(), GaussianNB())
pipe.fit(x_train, y_train)
y_pred = pipe.predict(x_test)
print(classification_report(y_test,y_pred))
precision = precision_score(y_test, y_pred,average='weighted')
print('Precision: %f' % precision)
# recall: tp / (tp + fn)
recall = recall_score(y_test, y_pred,average='weighted')
print('Recall: %f' % recall)
# f1: tp / (tp + fp + fn)
f1 = f1_score(y_test, y_pred,average='weighted')
print('F1 score: %f' % f1)

              precision    recall  f1-score   support

           0       0.00      0.00      0.00        18
           1       0.00      0.00      0.00         3
           2       0.00      0.00      0.00         1
           3       0.00      0.00      0.00         4
           4       0.08      0.40      0.14         5
           5       0.00      0.00      0.00         6
           6       0.00      0.00      0.00         6
           7       0.00      0.00      0.00         2
           8       0.08      0.14      0.11         7
           9       0.00      0.00      0.00         2
          12       0.00      0.00      0.00        11
          13       0.00      0.00      0.00         6
          14       0.00      0.00      0.00        15
          15       0.00      0.00      0.00         2
          16       0.10      0.43      0.16         7
          17       0.00      0.00      0.00         5

    accuracy                           0.06       100
   macro avg       0.02      0.06      0.03       100
weighted avg       0.02      0.06      0.03       100

Precision: 0.017000
Recall: 0.060000
F1 score: 0.025616

c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))


df = pd.DataFrame(classification_report(y_test,y_pred, output_dict=True)).transpose()
plt.figure()
df[['precision','recall','f1-score']].iloc[0:19].plot(kind='bar',figsize=(16,10))
plt.legend(loc='best')
plt.title("Classification Report Plot")

c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))

Text(0.5, 1.0, 'Classification Report Plot')

<Figure size 432x288 with 0 Axes>


plot_multiclass_roc_advanced(pipe, x_test, y_test, n_classes=16, figsize=(16, 10))


pipe = make_pipeline(StandardScaler(), RandomForestClassifier())
pipe.fit(x_train, y_train)
y_pred = pipe.predict(x_test)
print(classification_report(y_test,y_pred))
precision = precision_score(y_test, y_pred,average='weighted')
print('Precision: %f' % precision)
# recall: tp / (tp + fn)
recall = recall_score(y_test, y_pred,average='weighted')
print('Recall: %f' % recall)
# f1: tp / (tp + fp + fn)
f1 = f1_score(y_test, y_pred,average='weighted')
print('F1 score: %f' % f1)

              precision    recall  f1-score   support

           0       0.31      0.78      0.44        18
           1       0.00      0.00      0.00         3
           2       0.00      0.00      0.00         1
           3       0.00      0.00      0.00         4
           4       0.00      0.00      0.00         5
           5       1.00      0.17      0.29         6
           6       0.00      0.00      0.00         6
           7       0.00      0.00      0.00         2
           8       0.00      0.00      0.00         7
           9       0.00      0.00      0.00         2
          10       0.00      0.00      0.00         0
          12       0.00      0.00      0.00        11
          13       0.25      0.33      0.29         6
          14       0.26      0.33      0.29        15
          15       0.00      0.00      0.00         2
          16       0.08      0.14      0.11         7
          17       0.00      0.00      0.00         5

    accuracy                           0.23       100
   macro avg       0.11      0.10      0.08       100
weighted avg       0.18      0.23      0.17       100

Precision: 0.176307
Recall: 0.230000
F1 score: 0.165772

c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall is ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))


df = pd.DataFrame(classification_report(y_test,y_pred, output_dict=True)).transpose()
plt.figure()
df[['precision','recall','f1-score']].iloc[0:19].plot(kind='bar',figsize=(16,10))
plt.legend(loc='best')

c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))

<matplotlib.legend.Legend at 0x1b7298e7d00>

<Figure size 432x288 with 0 Axes>


plot_multiclass_roc_advanced(pipe, x_test, y_test, n_classes=16, figsize=(16, 10))


x_train = x_train[0:100]
y_train = y_train[0:100]
x_test = x_test[0:10]
y_test = y_test[0:10]


pipe = make_pipeline(StandardScaler(), XGBClassifier())
pipe.fit(x_train, y_train)
y_pred = pipe.predict(x_test)
print(classification_report(y_test,y_pred))
precision = precision_score(y_test, y_pred,average='weighted')
print('Precision: %f' % precision)
# recall: tp / (tp + fn)
recall = recall_score(y_test, y_pred,average='weighted')
print('Recall: %f' % recall)
# f1: tp / (tp + fp + fn)
f1 = f1_score(y_test, y_pred,average='weighted')
print('F1 score: %f' % f1)

c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].
  warnings.warn(label_encoder_deprecation_msg, UserWarning)

[23:20:33] WARNING: C:/Users/Administrator/workspace/xgboost-win64_release_1.4.0/src/learner.cc:1095: Starting in XGBoost 1.3.0, the default evaluation metric used with the objective 'multi:softprob' was changed from 'merror' to 'mlogloss'. Explicitly set eval_metric if you'd like to restore the old behavior.
              precision    recall  f1-score   support

           0       0.00      0.00      0.00         1
           1       1.00      1.00      1.00         1
           3       0.00      0.00      0.00         0
           4       1.00      0.50      0.67         2
           5       0.00      0.00      0.00         1
           6       0.00      0.00      0.00         1
           7       0.00      0.00      0.00         0
           8       0.00      0.00      0.00         0
          13       0.00      0.00      0.00         0
          14       0.00      0.00      0.00         4
          16       0.00      0.00      0.00         0
          17       0.00      0.00      0.00         0

    accuracy                           0.20        10
   macro avg       0.17      0.12      0.14        10
weighted avg       0.30      0.20      0.23        10

Precision: 0.300000
Recall: 0.200000
F1 score: 0.233333

c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision is ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall is ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))


plot_multiclass_roc_advanced(pipe, x_test, y_test, n_classes=6, figsize=(16, 10))


df = pd.DataFrame(classification_report(y_test,y_pred, output_dict=True)).transpose()
plt.figure()
df[['precision','recall','f1-score']].iloc[0:19].plot(kind='bar',figsize=(16,10))
plt.legend(loc='best')
plt.title("Classification Report Plot")

c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
c:\users\tom\appdata\local\programs\python\python39\lib\site-packages\sklearn\metrics\_classification.py:1245: UndefinedMetricWarning: Recall and F-score are ill-defined and being set to 0.0 in labels with no true samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))

Text(0.5, 1.0, 'Classification Report Plot')

<Figure size 432x288 with 0 Axes>


def tif_gzip_to_png(tif_path):
    png_path = pathlib.Path(tif_path.replace('.tif.gz','.png'))
    tf = gzip.open(tif_path).read()
    img = imageio.imread(tf, 'tiff')
    imageio.imwrite(png_path, img)


def download_and_convert_tifgzip_to_png(url, target_path):    
    r = requests.get(url)
    f = io.BytesIO(r.content)
    tf = gzip.open(f).read()
    img = imageio.imread(tf, 'tiff')
    imageio.imwrite(target_path, img)


all_locations = dict({
    "Nucleoplasm": 0,
    "Nuclear membrane": 1,
    "Nucleoli": 2,
    "Nucleoli fibrillar center": 3,
    "Nuclear speckles": 4,
    "Nuclear bodies": 5,
    "Endoplasmic reticulum": 6,
    "Golgi apparatus": 7,
    "Intermediate filaments": 8,
    "Actin filaments": 9,
    "Focal adhesion sites": 9,
    "Microtubules": 10,
    "Mitotic spindle": 11,
    "Centrosome": 12,
    "Centriolar satellite": 12,
    "Plasma membrane": 13,
    "Cell Junctions": 13,
    "Mitochondria": 14,
    "Aggresome": 15,
    "Cytosol": 16,
    "Vesicles": 17,
    "Peroxisomes": 17,
    "Endosomes": 17,
    "Lysosomes": 17,
    "Lipid droplets": 17,
    "Cytoplasmic bodies": 17,
    "No staining": 18
})


def add_label_idx(df, all_locations):
    '''Function to convert label name to index
    '''
    df["Label_idx"] = None
    for i, row in df.iterrows():
        labels = row.Label.split(',')
        idx = []
        for l in labels:
            if l in all_locations.keys():
                idx.append(str(all_locations[l]))
        if len(idx)>0:
            df.loc[i,"Label_idx"] = "|".join(idx)
            
        print(df.loc[i,"Label"], df.loc[i,"Label_idx"])
    return df


public_hpa_df = pd.read_csv('kaggle_2021.tsv')
# Remove all images overlapping with Training set
public_hpa_df = public_hpa_df[public_hpa_df.in_trainset == False]

# Remove all images with only labels that are not in this competition
public_hpa_df = public_hpa_df[~public_hpa_df.Label_idx.isna()]

colors = ['blue', 'red', 'green', 'yellow']
celllines = ['A-431', 'A549', 'EFO-21', 'HAP1', 'HEK 293', 'HUVEC TERT2', 'HaCaT', 'HeLa', 'PC-3', 'RH-30', 'RPTEC TERT1', 'SH-SY5Y', 'SK-MEL-30', 'SiHa', 'U-2 OS', 'U-251 MG', 'hTCEpi']
public_hpa_df_17 = public_hpa_df[public_hpa_df.Cellline.isin(celllines)]
len(public_hpa_df), len(public_hpa_df_17)

(77668, 67462)


public_hpa_df.head()


public_hpa_df.tail()


public_hpa_df[0:100]


labels = [str(i) for i in range(19)]
code = [
'00_Nucleoplasm',
'01_Nuclear_membrane',
'02_Nucleoli',
'03_Nucleoli_fibrillar_center',
'04_Nuclear_speckles',
'05_Nuclear_bodies',
'06_Endoplasmic_reticulum',
'07_Golgi_apparatus',
'08_Intermediate_filaments',
'09_Actin_filaments',
'10_Microtubules',
'11_Mitotic_spindle',
'12_Centrosome',
'13_Plasma_membrane',
'14_Mitochondria',
'15_Aggresome',
'16_Cytosol',
'17_Vesicles_and_punctate_cytosolic_patterns',
'18_Negative'
]

"""
for direct in code:
    save_dir = os.path.join(os.getcwd(),'data_img_2021/' + direct)
    if not os.path.exists(save_dir):
        os.makedirs(save_dir)
        
for label in labels:
    save_dir = os.path.join(os.getcwd(),'data_img_2021/' + code[int(label)])
    df_temp = public_hpa_df[public_hpa_df["Label_idx"] == label].reset_index(drop=True)
    dir_name = code[int(label)]
    for i, row in df_temp[0:3].iterrows():
        try:
            img = row.Image
            for color in colors:
                img_url = f'{img}_{color}.tif.gz'
                save_path = os.path.join(save_dir,  f'{os.path.basename(img)}_{color}.png')
                download_and_convert_tifgzip_to_png(img_url, save_path)
                print(f'Downloaded {img_url} as {save_path}')    
        except:
            print(f'failed to download: {img}')
"""

'\nfor direct in code:\n    save_dir = os.path.join(os.getcwd(),\'data_img_2021/\' + direct)\n    if not os.path.exists(save_dir):\n        os.makedirs(save_dir)\n        \nfor label in labels:\n    save_dir = os.path.join(os.getcwd(),\'data_img_2021/\' + code[int(label)])\n    df_temp = public_hpa_df[public_hpa_df["Label_idx"] == label].reset_index(drop=True)\n    dir_name = code[int(label)]\n    for i, row in df_temp[0:3].iterrows():\n        try:\n            img = row.Image\n            for color in colors:\n                img_url = f\'{img}_{color}.tif.gz\'\n                save_path = os.path.join(save_dir,  f\'{os.path.basename(img)}_{color}.png\')\n                download_and_convert_tifgzip_to_png(img_url, save_path)\n                print(f\'Downloaded {img_url} as {save_path}\')    \n        except:\n            print(f\'failed to download: {img}\')\n'


# ! pip install https://github.com/CellProfiling/HPA-Cell-Segmentation/archive/master.zip


import hpacellseg.cellsegmentator as cellsegmentator
from hpacellseg.utils import label_cell, label_nuclei
import glob
code = [
'00_Nucleoplasm',
'01_Nuclear_membrane',
'02_Nucleoli',
'03_Nucleoli_fibrillar_center',
'04_Nuclear_speckles',
'05_Nuclear_bodies',
'06_Endoplasmic_reticulum',
'07_Golgi_apparatus',
'08_Intermediate_filaments',
'09_Actin_filaments',
'10_Microtubules',
'11_Mitotic_spindle',
'12_Centrosome',
'13_Plasma_membrane',
'14_Mitochondria',
'15_Aggresome',
'16_Cytosol',
'17_Vesicles_and_punctate_cytosolic_patterns',
'18_Negative'
]
directorys_lst = os.listdir('data_img_2021/')
label_tracker = 0

for q  in range(19):
    save_dir = 'data_img_2021/'+directorys_lst[q]
    
    mt = glob.glob(save_dir + '/' + '*_red.png')
    er = [f.replace('red', 'yellow') for f in mt]
    nu = [f.replace('red', 'blue') for f in mt]
    images = [mt, er, nu]
    print(images)
  
    NUC_MODEL = "./nuclei-model.pth"
    CELL_MODEL = "./cell-model.pth"
    segmentator = cellsegmentator.CellSegmentator(
        NUC_MODEL,
        CELL_MODEL,
        scale_factor=0.25,
        device="cuda",
        padding=True,
        multi_channel_model=True,
    )

    # For nuclei
    nuc_segmentations = segmentator.pred_nuclei(images[2])

    # For full cells
    cell_segmentations = segmentator.pred_cells(images)

    # post-processing
    for i, pred in enumerate(cell_segmentations):
        nuclei_mask, cell_mask = label_cell(nuc_segmentations[i], cell_segmentations[i])
        FOVname = os.path.basename(mt[i]).replace('red','predictedmask')
        imageio.imwrite(os.path.join(save_dir,FOVname), cell_mask) 
    
    print("Generating masking images for "+ code[label_tracker])
    if not os.path.exists("data_img/"+directorys_lst[q]):
        os.makedirs("data_img/"+directorys_lst[q])
    saved_dir_last = "data_img/"+directorys_lst[q]


    df_temp = public_hpa_df[public_hpa_df["Label_idx"] == str(label_tracker)].reset_index(drop=True)
    lst = list()
    for i, row in df_temp[0:3].iterrows():
        img = row.Image
        lst.append(img.split('/')[4])
    
    # Visualizing the segmentation masks we just predicted above
    for i, row in df_temp[0:3].iterrows():
        fig, ax = plt.subplots(figsize=(180,180))
        img = row.Image

        microtubule = plt.imread(mt[i])    
        endoplasmicrec = plt.imread(er[i])    
        nuclei = plt.imread(nu[i])

        mask = plt.imread(mt[i].replace('red','predictedmask'))
        img = np.dstack((microtubule, endoplasmicrec, nuclei))

        ax.imshow(img)
        ax.imshow(mask, alpha=0.5)
        ax.axis('off')


        plt.savefig(saved_dir_last+'/'+lst[i] , bbox_inches='tight')
        
        
    print("Completed Generating masking images for "+ code[label_tracker])
    label_tracker+=1


train_ds = tf.keras.preprocessing.image_dataset_from_directory(
 "./data_img",
  validation_split=0.2,
  subset="training",
  seed=123,
  image_size=(img_height, img_width),
  batch_size=batch_size)

Found 56 files belonging to 19 classes.
Using 45 files for training.


class_names = train_ds.class_names
print(class_names)

['00_Nucleoplasm', '01_Nuclear_membrane', '02_Nucleoli', '03_Nucleoli_fibrillar_center', '04_Nuclear_speckles', '05_Nuclear_bodies', '06_Endoplasmic_reticulum', '07_Golgi_apparatus', '08_Intermediate_filaments', '09_Actin_filaments', '10_Microtubules', '11_Mitotic_spindle', '12_Centrosome', '13_Plasma_membrane', '14_Mitochondria', '15_Aggresome', '16_Cytosol', '17_Vesicles_and_punctate_cytosolic_patterns', '18_Negative']


plt.figure(figsize=(20, 20))
for images, labels in train_ds.take(1):
    for i in range(19):
        ax = plt.subplot(4, 5, i + 1)
        plt.imshow(images[i].numpy().astype("uint8"))
        plt.title(class_names[labels[i]])
        plt.axis("off")

	ID	Label
0	5c27f04c-bb99-11e8-b2b9-ac1f6b6435d0	8\|5\|0
1	5fb643ee-bb99-11e8-b2b9-ac1f6b6435d0	14\|0
2	60b57878-bb99-11e8-b2b9-ac1f6b6435d0	6\|1
3	5c1a898e-bb99-11e8-b2b9-ac1f6b6435d0	16\|10
4	5b931256-bb99-11e8-b2b9-ac1f6b6435d0	14\|0

	ID	ImageWidth	ImageHeight	PredictionString
0	0040581b-f1f2-4fbe-b043-b6bfea5404bb	2048	2048	0 1 eNoLCAgIMAEABJkBdQ==
1	004a270d-34a2-4d60-bbe4-365fca868193	2048	2048	0 1 eNoLCAgIMAEABJkBdQ==
2	00537262-883c-4b37-a3a1-a4931b6faea5	2048	2048	0 1 eNoLCAgIMAEABJkBdQ==
3	00c9a1c9-2f06-476f-8b0d-6d01032874a2	2048	2048	0 1 eNoLCAgIMAEABJkBdQ==
4	0173029a-161d-40ef-af28-2342915b22fb	3072	3072	0 1 eNoLCAgIsAQABJ4Beg==

	Image	Label	Cellline	in_trainset	Label_idx
0	https://images.proteinatlas.org/10005/921_B9_1	Cytosol	A-431	False	16
1	https://images.proteinatlas.org/10005/921_B9_2	Cytosol	A-431	False	16
2	https://images.proteinatlas.org/10005/923_B9_1	Cytosol	U-251 MG	False	16
3	https://images.proteinatlas.org/10005/923_B9_2	Cytosol	U-251 MG	False	16
4	https://images.proteinatlas.org/10005/931_B9_1	Cytosol	U-2 OS	False	16

	Image	Label	Cellline	in_trainset	Label_idx
82490	https://images.proteinatlas.org/9985/68_B8_2	Cytosol	U-251 MG	False	16
82491	https://images.proteinatlas.org/9985/69_B8_1	Cytosol	A-431	False	16
82492	https://images.proteinatlas.org/9985/69_B8_2	Cytosol	A-431	False	16
82493	https://images.proteinatlas.org/9985/91_B8_1	Mitochondria,Cytosol	U-2 OS	False	14\|16
82494	https://images.proteinatlas.org/9985/91_B8_2	Mitochondria,Cytosol	U-2 OS	False	14\|16

Group Members: Michael Woo, Thomas Slawinski, Akaash Patel, Rahul Patel¶

Imports¶

Exploring the data¶

Note¶

Number of labels per images¶

Things to keep in mind regarding channels¶

Here we seperate the images based on the ID and color¶

Viewing the images¶

Generating Stacked Images (Feature Engineering)¶

0. Nucleoplasm

1. Nuclear membrane

2. Nucleoli

3. Nucleoli fibrillar center

4. Nuclear speckles

5. Nuclear bodies

6. Endoplasmic reticulum

7. Golgi apparatus

8. Intermediate filaments

9. Actin filaments

10. Microtubules

11. Mitotic spindle

12. Centrosome

13. Plasma membrane

14. Mitochondria

15. Aggresome

16. Cytosol

17. Vesicles and punctate cytosolic patterns

18. Negative

Generating directories and saving images with respect to the labels associated to the images¶

Further Feature Engineering¶

Size of images and batch size variables¶

Training Data Generator¶

Validation Data Generator¶

Training Data¶

Validation data¶

Subsetting Data¶

Create functions to create multi-label AUC_ROC curve¶

Naive Bayes Classifer¶

Plot classification report results (Naive Bayes)¶

Random Forest Classifer¶

Plot classification report results (Random Forest)¶

XGBoost Classifer¶

Reducing subset specifically for XGBoost to speed up model training¶

Plot classification report results (XGBoost)¶

Cellular Masking (further potential feature engineering)¶

Idea: Mask images by color to generate more easily readble images to provide to the models as input¶

Didn't perform due to limited computational resources and time when attempting to use a realistic sample size (3 stacked images per classification took almost an hour)¶

Function to convert .tif.gz to .png and put it in the same folder¶

Function to convert .tif.gz to .png and put it in the same folder¶

All label names in the public HPA and their corresponding index¶

Function to convert label name to index¶

Reading in TSV file¶

Only have to install this once¶

Segmenting the cells¶

Visualization of images stacked with masking¶